Applying Points-of-View Analysis to Individual Variations in Colour Sorting Data

Abstract

“Points-of-View” analysis has been promoted as an appropriate analysis for similarity data collected with the Method of Sorting. It can be regarded as an extension of Cultural Consensus Analysis (cca). The latter assumes that subjects all base their responses on a single shared ‘model’ of the items to be sorted (while varying in the level of reliability with which they consult that model). Conversely, the titular “points-of-view” are multiple models, sampled singularly by some subjects’ responses, while other subjects combine the models in various proportions. The analysis appears to be comparatively insensitive to the artefacts to which sorting data are prone, which affect how easily they can be interpreted with (for instance) multidimensional scaling (mds). Here we apply the Points-of-View approach to two sets of data from the colour domain, a sensory modality well-suited for the sorting procedure. One study extracted three viewpoints – i.e., factors, prototypal ways of organising the stimuli – with one viewpoint specific to colour-vision-deficient observers, capturing independently-measured differences among the subjects. In the second study, two viewpoints proved to be appropriate: one arranged the items by two dimensions, lightness/saturation as well as hue, while the second was essentially a one-dimensional hue-based arrangement. This distinction could not be recovered from the raw data by applyingmdswith the weighted-Euclidean model of individual difference. We discuss some demographic factors which might dispose subjects to attend to hue only or saturation/lightness as well.